Hierarchical clustering of samples using phylum level distributions. Employing an alternative Qiime-based methodology to analyze our sequences, we see that water samples consistently cluster within their own specific environments. Again, this is not so for the fruit surface samples. Displayed values are log transformed relative abundances within each sample, (e.g. 0.10 ~-1; 0.01 ~-2). Visualized using skiff in CloVR.